International Conference on Electrical, Electronics, and Optimization Techniques (ICEEOT) - 2016 Analytical review of three latest nature inspired algorithms for scheduling in clouds Navneet Kaur Computer Engineering and Technology GNDU Amritsar, India navigr892@gmail.com Abstract-Cloud computing is one of the current research topics because of its increasing demand and fancy features like pay as you go facility, resource pooling, rapid elasticity, on demand services and broad network access. Since managing a large pool of resources involved in a cloud computing scenario is difficult as well as a necessary job, effective scheduling techniques must be there to tackle with the problem. This paper reviews three recently developed nature inspired metaheuristic techniques namely firefly algorithm, cuckoo search and bat algorithm for dealing with scheduling problem, discuss their advantageous features and the future scope of the algorithms for scheduling in computing environment of clouds. Keywords-cloud computing; metaheuristics; firefly algorithm; cuckoo search; bat algorithm. I. INTRODUCTION Scheduling in clouds is an NP-Hard problem and finding an optimal solution is a challenge. Deterministic methods like gradient based algorithms work well for continuous and unimodal problems. If applied to scheduling in clouds, these algorithms may not provide satisfactory solutions because of large search space and discrete nature of the problem. Metaheuristic techniques can efficiently tackle these issues and can find near-optimal solutions in comparatively shorter period of time. Metaheuristic refers to a high level procedure which aims at finding, generating or selecting heuristic that can provide an approximate solution to an optimization problem [1]. This technique can efficiently solve problems that are larger in size and are complex. Sufficient research has already been done regarding older metaheuristic techniques like genetic algorithms (GA), ant colony optimization (ACO), particle swarm optimization (PSO) and their application in the field of cloud scheduling. Various nature inspired metaheuristic techniques have been developed in the recent past and are performing better than the above mentioned older methods. This paper reviews three recent nature inspired metaheuristic methods namely firefly algorithm, cuckoo search and bat algorithm that have been proved very efficient in various fields and have tremendous scope in the field of scheduling in clouds too. Amit Chhabra Computer Engineering and Technology GNDU Amritsar, India amit.cse@gndu.ac.in II. FIREFLY ALGORITHM Firefly algorithm (FA) is based on the flashing behaviour of the tropical fireflies. Xin-She Yang had developed the algorithm in 2008 [2,3]. Following characteristics of the fireflies are used for the development of FA. The fireflies will be attracted to each other regardless of their sex. Brighter is the flash, more will be the attractiveness and with distance, the attractiveness and brightness both decreases. The brightness of a firefly is affected or determined by the landscape of the objective function. If a problem is maximization problem, the brightness value of a firefly can be proportional to objective function's value. The pseudo code shown in Fig. 1 summarizes the basic steps of the firefly algorithm. Fig. 1. Pseudo code of the firefly algorithm. The discrete form of firefly algorithm was developed by Sayadi et al. [4] for its application to the scheduling problems. To deal with multi-objective optimization problem, multi-
objective firefly algorithm (MOFA) was developed by Yang [5] for continuous design problems which was further extended for discrete problems too by Apostolopoulos and Vlacho [6]. In grid computing, taking energy consumption as an optimization criteria, another multiobjective firefly algorithm was proposed by Arsuaga-Rias and Vega-Rodriguez independently [7]. FA was used for solving production scheduling problem with multiple objectives by Li & Ye et al. [8]. Then for dealing with multi-objective flowshop scheduling system, another variant of FA was presented by Marichelvam et al. [9]. Chaos as a parameter was also used that further enhanced the performance of FA [10,11]. Specifically for scheduling in clouds, a jumper firefly algorithm which aimed at reducing the makespan was used [12]. Furthermore, an improved firefly algorithm that used information communication of fireflies for finding global optimal solution for scheduling problem in clouds was developed [13]. Another firefly algorithm based on chaos optimization for solving resource scheduling problem in cloud computing was developed [14]. B. Efficient features Since firefly algorithm is based on swarm-intelligence, it already has advantages that swarm based algorithms have. Other major advantages of FA are the automatic subdivision of the population and feature of dealing with multimodality. The entire population subdivides into smaller groups automatically and these subgroups swarm around each of the local optimum. Also, all optima are found simultaneously when population size is high. Not only this, parameter tuning is also done which controls the randomness which in turn speeds up the convergence. III. CUCKOO SEARCH ALGORITHM Brood parasitism based algorithm, cuckoo search was developed by Xin-She Yang and Swash Deb in 2009 [15,16,17]. A standard cuckoo search algorithm is described by following rules : One egg is laid by the cuckoo at a time and dumps it in a nest which is chosen randomly. the nest which has better quality eggs will only be carried further to the next generation. There are fixed number of host nests and the discovery of the egg laid by cuckoo by the host bird is with probability p a. The egg can either be discarded by the host bird or the host bird can discard the nest and build a new one. Here a solution is represented by the bird cuckoo which is assumed to lay only one egg. The algorithm concentrates at replacing weaker solutions with better ones. The algorithm balances both local and global random walks by controlling p a which is a switching parameter. The pseudo code of the algorithm is described in Fig. 2. Fig. 2. Pseudo code of cuckoo search algorithm. Few variants were made in order to build discrete form of cuckoo search algorithm. For dealing with muti objective optimization, MOCS i.e multi objective cuckoo search was developed by Yang and Deb et al. [17]. For scheduling purpose, discrete form of MOCS was presented by Chandrasekaran and Simon [18]. For job shop scheduling problem, discrete cuckoo search algorithm was applied [19]. Another variant named as modified cuckoo search which was developed by Walten et al [20] was used for flow shop scheduling for minimising the makespan [21]. For scheduling in clouds, a hybridized cuckoo search algorithm which combined the ACO features with cuckoo search algorithm was made aiming at minimising the completion time [22]. A hybrid of dynamic particle swarm algorithm and cuckoo search was devised for scheduling resources in a cloud environment by optimizing execution time [23]. Another attempt to apply cuckoo search algorithm for task scheduling in clouds were made in [24]. B. Efficient Features Cuckoo search algorithm does not use simple random walks instead it uses Levy flights [25]. Due to infinite mean and variance of levy flights, this algorithm can efficiently explore the entire search space. Also cuckoo search algorithm ensures the global convergence property. As compared to particle swarm optimization algorithms, which can converge to local optima prematurely, cuckoo search algorithm usually converges to global optimal solutions. Not only this, cuckoo search algorithm can effectively balance the local and global search with the help of switching parameter. The local search takes about one fourth of the total execution time and global search about three fourth of the total search time with the value of the switching parameter as 0.25. This allows to achieve global optimality with higher probability.
IV. BAT ALGORITHM Bat algorithm (BA) developed by Xin-She Yang in the year 2010 is based on the echolocation behaviour of the virtual bats [26]. Following rules are used for framing the bat algorithm: Echolocation is used by bats to measure the distance of food/prey and the difference between food and the environmental barriers is well known to them. Bats fly with certain velocity at a particular position where velocity and position are represented by v i and x i respectively. one of the two factors, frequency and wavelength, is kept fixed and the other one is adjusted automatically by the bats and parameters named pulse rate and loudness are adjusted depending upon the distance of the target. Pulse rate is denoted by r and vary in the interval [0,1] whereas loudness is assumed to vary from a large positive value denoted by A 0 to a minimum value denoted by A min.. The pseudo code for the algorithm is given below in Fig. 3. Binary bat algorithm was used by S Raghavan aiming at minimizing the cost factor [31]. B. Efficient features BA has many efficient features. The first one is frequency tuning. Frequency variation in BA enables the key features of algorithms like PSO, simulated annealing to be present in the algorithm. The second feature is automatic zooming into the region of high quality solutions. BA also ensures balance between exploration and exploitation due to this feature. Convergence speed of BA is also high as compared to other methods. Other important feature of BA is parameter control. Variation in parameters namely loudness and emission rate helps in switching automatically from exploration to exploitation when optimal global solution is approaching near. BA is highly efficient algorithm that can effectively solve large scale problems and also guarantee quick convergence to global solution under appropriate conditions. The comparison of three on the basis of the technique used, parameters used, features and work done in the field of scheduling in clouds is shown in Table 1. V. DISCUSSION For managing the cloud resources to ensure smooth working of cloud system, effective scheduling is very important. Metaheuristic algorithms like ACO, PSO and GA have widely been applied for this purpose with many variations but the latest nature inspired metaheuristic algorithms discussed above namely firefly algorithm, cuckoo search and bat algorithm, despite of being very efficient are merely explored for the purpose of scheduling in clouds. Since almost every organisation is now using cloud system, need of efficient scheduling algorithm is increasing, further exploration of these latest algorithms can definitely give better results as compared to the old metaheuristic methods. Fig. 3. Pseudo code of BAT algorithm. Binary bat algorithm (BBA) which is the discrete version of bat algorithm was devised by Nakamura et al. [27]. Multi objective bat algorithm (MOBA) is the extended version of BA to deal with multi objective problems. A production scheduling problem including multiple stages, machines and products was solved by Musikapun and Pongcharoen in 2012 using BA and also suggested that with better choice of parameters, performance can improve by 8.4% approximately [28]. With execution time and mean flow time as optimization criteria, Marichelvam and Prabaharan used BA to solve flow shop scheduling problems [29]. For scheduling in clouds, BA was used by jacob for resource scheduling with makespan as optimization criteria [30]. A hybrid algorithm comprising PSO and MOBA was devised for maximizing profit in clouds. Attention should be paid on following research areas and challenges which will improve the scope of these new algorithms to be used for the purpose of scheduling in clouds. Working on issues of convergence speed, time complexity and stability of the new nature inspired methods will make more efficient for use in solving scheduling problem in clouds. Intense research and study needs to be done regarding statistical measures, performance analysis and comparison of these new metaheuristic algorithms with other algorithms. Working in the area of parameter control and tuning of these algorithms would maximize their performance and further increase the chances of producing high quality results when applied to the field of scheduling in clouds.
Intelligently developing hybrid algorithms can improve the quality of solutions. This requires deep understanding of basic components of algorithm, experience and expertise of the developer. Focus should be on components like crossover, mutation, random walks, chaos, levy flights and gradients. These latest algorithms are almost new to the field of scheduling in clouds. Very less work has been done in this field and only execution time as an optimization criteria is chosen. To fit better with cloud environment, fitness function must be modified. Factors other than makespan like cost, load balancing, energy efficiency, communication cost, reliability etc must also be included for optimization by these new algorithms. TABLE I. COMPARISON OF FIREFLY, CUCKOO SEARCH ALGORITHM AND BAT ALGORITHM. NAME OF THE METHOD FIREFLY ALGORITHM BASIS PARAMETERS USED ADVANTAGEOUS FEATURES flashing behaviour and flashing pattern of the fireflies Randomization parameter Light absorption coefficient Maximum number of generations 1. Automatic subdivision of entire population. 2. Efficiently deal with multimodality. 3.Increases convergence speed. WORK DONE IN THE FIELD OF SCHEDULING IN CLOUD COMPUTING ENVIRONMENT [12], [13],[14] CUCKOO SEARCH property of brood parasitism of cuckoo speices Switching probability Step-size scaling factor Levi exponent 1. Ensures the property of global convergence due to the use of switching probability factor. 2. Use of levy flights result in efficient exploration of search space. [22], [23], [23] BAT ALGORITHM echolocation behaviour of virtual bats. Loudness parameter Pulse rate Maximum number of iterations 1. Frequency tuning 2. Automatic zooming into the region of global solution [29], [30], [31] 3. Parameter control ensures efficient exploration and exploitation.
International Conference on Electrical, Electronics, and Optimization Techniques (ICEEOT) - 2016 VI. CONCLUSION The paper reviews latest nature inspired algorithms namely firefly algorithm, cuckoo search and bat algorithm for scheduling in cloud computing environment. Paper discusses why these algorithms are efficient and how their further exploration in this field can yield good results. Comparison of the three algorithms have been made on the basis of parameters used, efficiency features and work done in the field of cloud resource scheduling. These algorithms have already been widely used in areas like optimization, image processing, data mining, feature selection, classification including scheduling and many more. The tremendous results shown by them in these fields and their efficient features clearly indicate that in future, use of these algorithms in the field of scheduling in cloud environment, will be very beneficial. References [1] Talbi eg Information, Wiley: Metaheuristics: From design to implementation - El-Ghazali Talbi. 2009. [2] X.-S. Yang, Nature-inspired Metaheuristic Algorithms, 1st ed. Frome, UK: Luniver Press, 2008. [3] X.-S. Yang, Firefly algorithms for multimodal optimization, in: Stochastic Algorithms: Foundations and Applications, SAGA 2009, Lecture Notes in Computer Sciences, Vol. 5792, pp. 169-178 (2009). [4] M. K. Sayadi, R. Ramezanian, and N. Ghaffari-Nasab, "A discrete firefly meta-heuristic with local search for makespan minimization in permutation flow shop scheduling problems," International Journal of Industrial Engineering Computations, vol. 1, no. 1, pp. 1 10, Jul. 2010. [5] X.-S. Yang, "Multiobjective firefly algorithm for continuous optimization," Engineering with Computers, vol. 29, no. 2, pp. 175 184, Jan. 2012. [6] T. Apostolopoulos and A. Vlachos, "Application of the Firefly algorithm for solving the economic emissions load dispatch problem," International Journal of Combinatorics, vol. 2011, pp. 1 23, 2011. [7] M. Arsuaga-Ríos and M. A. Vega-Rodríguez, Multi-objective Firefly algorithm for energy optimization in grid environments in Lecture Notes in Computer Science. Springer Science + Business Media, 2012, pp. 350 351. [8] C. Ye and H. Li, "Firefly algorithm on multi-objective optimization of production scheduling system," Advances in Mechanical Engineering and its Applications, vol. 3, no. 1, pp. 258 262, Oct. 2012. [9] M. K. Marichelvam, T. Prabaharan, and X. S. Yang, "A discrete Firefly algorithm for the multi-objective hybrid Flowshop scheduling problems," IEEE Transactions on Evolutionary Computation, vol. 18, no. 2, pp. 301 305, Apr. 2014. [10] L. dos Santos Coelho, D. L. de Andrade Bernert, and V. C. Mariani, "A chaotic firefly algorithm applied to reliabilityredundancy optimization," 2011 IEEE Congress of Evolutionary Computation (CEC), Jun. 2011. [11] X.-S. Yang, "Chaos-enhanced Firefly algorithm with automatic parameter tuning," International Journal of Swarm Intelligence Research, vol. 2, no. 4, pp. 1 11, 2011. [12] G. Nithya and R. M. S. Engels, "Multi-agent brokering approach and firefly algorithm for job scheduling in grid environment," 2014 International Conference on Electronics and Communication Systems (ICECS), Feb. 2014. [13] L. F. Zhao, S. H. Zhou, W. B. Chang, "Task Scheduling in Cloud Computing with Improved Firefly Algorithm", Applied Mechanics and Materials, Vols. 602-605, pp. 3189-3193, Aug. 2014. [14] Y. Miao, "Resource scheduling simulation design of Firefly algorithm based on chaos optimization in cloud computing,"international Journal of Grid and Distributed Computing, vol. 7, no. 6, pp. 221 228, Dec. 2014. [15] X.-S. Yang, S. Deb, Cuckoo search via L evy flights, in: Proc. of World Congress on Nature & Biologically Inspired Computing (NaBIC 2009), December 2009, India. IEEE Publications, USA, pp. 210-214 (2009). [16] X. S. Yang and S. Deb, "Engineering optimisation by cuckoo search," International Journal of Mathematical Modelling and Numerical Optimisation, vol. 1, no. 4, p. 330, 2010. [17] X.-S. Yang and S. Deb, "Multiobjective cuckoo search for design optimization," Computers & Operations Research, vol. 40, no. 6, pp. 1616 1624, Jun. 2013. [18] K. Chandrasekaran and S. P. Simon, "Multi-objective scheduling problem: Hybrid approach using fuzzy assisted cuckoo search algorithm," Swarm and Evolutionary Computation, vol. 5, pp. 1 16, Aug. 2012 [19] A. Ouaarab, B. Ahiod, X.-S. Yang, and M. Abbad, "Discrete cuckoo search algorithm for job shop scheduling problem," 2014 IEEE International Symposium on Intelligent Control (ISIC), Oct. 2014. [20] S. Walton, O. Hassan, K. Morgan, and M. R. Brown, "Modified cuckoo search: A new gradient free optimisation algorithm," Chaos, Solitons & Fractals, vol. 44, no. 9, pp. 710 718, Sep. 2011. [21] H. Wang, W. Wang, H. Sun, C. Li, S. Rahnamayan, and Y. Liu, "A modified cuckoo search algorithm for flow shop scheduling problem with blocking," 2015 IEEE Congress on Evolutionary Computation (CEC), pp. 456 463, May 2015. [22] R. Raju, R. G. Babukarthik, D. Chandramohan, P. Dhavachelvan, and T. Vengattaraman, "Minimizing the makespan using hybrid algorithm for cloud computing," 2013 3rd IEEE International Advance Computing Conference (IACC), pp. 957 96, Feb. 2013. [23] A. Al-maamari and F. A. Omara, "Task scheduling using PSO algorithm in cloud computing environments," International Journal of Grid and Distributed Computing, vol. 8, no. 5, pp. 245 256, Oct. 2015. [24] N. Jafari Navimipour and F. Sharifi Milani, "Task scheduling in the cloud computing based on the cuckoo search algorithm,"international Journal of Modeling and Optimization, vol. 5, no. 1, pp. 44 47, Feb. 2015. [25] Pavlyukevich, "Lévy flights, non-local search and simulated annealing," Journal of Computational Physics, vol. 226, no. 2, pp. 1830 1844, Oct. 2007. [26] X.-S. Yang, A new Metaheuristic bat-inspired algorithm innature Inspired Cooperative Strategies for Optimization (NICSO 2010). Springer Science + Business Media, 2010, pp. 65 74. [27] R. Y. M. Nakamura, L. A. M. Pereira, K. A. Costa, D. Rodrigues, J. P. Papa, and X.-S. Yang, "BBA: A binary bat algorithm for feature selection," 2012 25th SIBGRAPI Conference on Graphics, Patterns and Images, Aug. 2012. [28] P. Musikapun and P. Pongcharoen, "Solving multi-stage multimachine multi-product scheduling problem using bat algorithm,"second international conference on management and artificial intelligence (IPEDR), vol. 35, pp. 98 102, 2012 [29] M. K. Marichelvam and T. Prabaharan, "A bat algorithm for realistic hybrid flowshop scheduling problems to minimize makespan and mean flow time," ICTACT Journal on Soft Computing, vol. 3, no. 1, Oct. 2012. [30] L. Jacob, "Bat algorithm for resource scheduling in cloud computing," International Journal For Research In Applied Science And Engineering Technology (IJRASET), vol. 2, no. 4, 2014 [31] S. Raghavan, P. Sarwesh, C. Marimuthu, and K. Chandrasekaran, "Bat algorithm for scheduling workflow applications in cloud," 2015 International Conference on Electronic Design, Computer Networks & Automated Verification (EDCAV), Jan. 2015.